Fully Automated Radiomic Analysis to Differentiate Pain Profiles in Femoroacetabular Impingement: A Robust MRI-Based Model

Motivation: This study aims to enhance MRI-based differentiation of pain profiles in femoroacetabular impingement (FAI) for improved treatment planning. Goal(s): Develop a robust radiomic model to distinguish symptomatic, asymptomatic, and healthy hips in FAI patients. Approach: We used MRI data from three cohorts, including an external validation set of over 185 independently acquired patients, with automated segmentation, radiomic feature extraction, and machine learning to classify hip states. Results: The models achieved >85% accuracy, with some classifiers reaching 100%, demonstrating reliability and generalizability across varied imaging protocols and independently acquired data. Impact: This fully automated method allows clinicians to reliably differentiate pain profiles in FAI patients across diverse imaging protocols, enhancing personalized treatment strategies. It paves the way for broader clinical adoption of MRI-based radiomics for musculoskeletal disorders, supporting efficient, reproducible diagnostics.